Introduction
Gastric cancer is a significant medical disease worldwide, ranking fifth in cancer
incidence and third in mortality around the globe [1]. When diagnosed and treated at an early stage, gastric cancer carries a significantly
better overall survival. In countries where a high proportion of early gastric cancers
are detected, a higher incidence:mortality ratio is observed. Unfortunately, the proportion
of cases diagnosed early remains low in most areas outside of Japan and Korea. A study
reporting outcomes in patients undergoing gastrectomy for gastric cancers found that
only 205 of cancers were found at an early stage in the United States and versus 50 %
in Japan [2]. Population screening programs are well established in countries with the highest
incidence of cancer, allowing for earlier cancer detection. In medium to low incidence
regions, population screening is not a cost-effective strategy. Screening and surveillance
of high-risk individuals has been recommended in the latest European guidelines [3].
Atrophic gastritis (AG) and intestinal metaplasia (IM) are established risk factors
for gastric carcinogenesis [4]. However, there is currently lack of a standard in documenting the degree of atrophy
and intestinal metaplasia. Endoscopic grading and histological grading have both been
used occasionally in different countries [5]
[6]
[7]
[8]
[9]
[10]. In this study, we aimed to investigate current practice and training in terms of
endoscopic assessment of AG and IM.
Methods
The study was designed as an online survey using Internet software Google Form (https://goo.gl/forms/nHqusNiZsHSn72eN2). The questionnaire consisted of 1) basic demographics; 2) endoscopists’ current
situation in diagnosis and characterization of AG and IM; and 3) a an image test for
endoscopic diagnosis of AG and IM.
Information regarding participants’ basic demographics was collected, including age,
gender, type of hospital, years of experience as a medical doctor, number of diagnostic
endoscopies per week, and country of practice. Data on endoscopists’ current practice
on diagnosis and characterization of AG and IM, reporting of endoscopic findings,
classification system used for AG, reporting on distribution of IM and biopsy protocol
for diagnosis of AG or IM also were collected. Specific questions were asked about
the endoscopists’ prior education on diagnosis and characterization of AG and IM.
Finally, an image test consisting of seven cases (Helicobacter pylori-naïve normal patients, n = 3; patients with AG and IM, n = 4) with endoscopic images
of the stomach was used to assess participants’ ability to recognize the severity
of AG and IM. Examples of the questions are shown in [Fig. 1] and [Fig. 2]. Three questions were about the diagnosis and grading of gastric atrophy and four
questions were about the diagnosis and grading of IM ([Fig. 3]). In the four questions on IM, endoscopic images with narrow band imaging were provided
for the participants’ assessment. Two of the cases had severe AG visualized on the
still images by pale-colored mucosa, increased visibility of submucosal vasculatures,
and loss of gastric folds that involved the entire corpus of the stomach including
the lesser and greater curvature. Two of the cases had significant IM involving both
the antrum and corpus, as observed in the still images with multiple ridged/tubular
epithelium and light blue crest sign in all parts of the stomach.
Fig. 1 Proportion of participants with prior training in AG or IM.
Fig. 2 Sample question 1 in the picture test. “What is the degree of atrophic gastritis?”
Answer – Severe atrophic gastritis (Open type II or III).
Fig. 3 Sample question 2. “What is the degree of intestinal metaplasia?” Answer – Extensive
intestinal metaplasia involving corpus and antrum.
The questionnaire was sent out to endoscopists around the world through e-mail by
the authors (H.Y. and N.U.). Recipients were encouraged to invite colleagues at any
level of training with prior experience in upper gastrointestinal endoscopy to participate
in the online survey.
Results from the online questionnaire were collected and analyzed. The difference
in endoscopic practice between endoscopists from different parts of the world was
assessed. The accuracy in endoscopic assessment of GA was also compared. In particular,
comparison was made between endoscopists in Japanese and Korea, where a high proportion
of early gastric cancers are detected, versus the rest of the participating endoscopists
around the world. Continuous data were analyzed with the student’s t-test or Mann Whitney U test while categorical data were analyzed with X2 test. Data analysis was performed with IBM SPSS statistics ver. 19.
Results
From July to October 2017, 249 endoscopists responded to the survey. Demographic information
on the participants is shown in [Table1]. The majority of them had > 15 years experience in endoscopy and worked in a university-affiliated
hospital. More than 75 % of the participants with the highest incidence of gastric
cancer came from Asia and South America.
Table 1
Background information on participating endoscopists.
|
Total
|
|
Age
|
|
|
20 (8.0 %)
|
|
|
37 (14.9 %)
|
|
|
53 (21.3 %)
|
|
|
139 (55.8 %)
|
|
Gender
|
|
|
190 (76.3 %)
|
|
|
59 (23.7 %)
|
|
Type of hospital at work
|
|
|
109 (43.8 %)
|
|
|
73 (29.3 %)
|
|
|
23 (9.2 %)
|
|
|
40 (16.1 %)
|
|
|
4 (1.6 %)
|
|
Years of experience
|
|
|
17 (6.8 %)
|
|
|
41 (16.5 %)
|
|
|
51 (20.5 %)
|
|
|
41 (16.5 %)
|
|
|
99 (39.8 %)
|
|
Number of diagnostic endoscopy per week
|
|
|
43 (17.3 %)
|
|
|
95 (38.2 %)
|
|
|
81 (32.5 %)
|
|
|
30 (12.0 %)
|
|
Countries of practice
|
|
|
34 (13.7 %)
|
|
|
56 (21.3 %)
|
|
|
32 (12.9 %)
|
|
|
16 (6.4 %)
|
|
|
105 (42.2 %)
|
|
|
6 (2.4 %)
|
|
|
3 (1.2 %)
|
In our study, 69.5 % (173/249) and 71.1 % (177/249) of endoscopists received prior
training regarding endoscopic diagnosis of AG and IM, respectively. Only 30.5 % of
them received such training within the first or second year of their endoscopic careesr.
A significant variation was observed regarding their training around the globe ([Fig. 2]).
Most endoscopists would routinely report the presence of AG and IM (70.7 %) ([Table 2]). The classification systems being used for documenting degree of AG differed in
different areas. More than 90 % of Japanese and Korean endoscopists would use the
Kimura-Takemoto classification while 51.4 % of South American and 43.8 % of European
would use the Operative Link for Gastritis Assessment (OLGA) system. Up to 56.6 %
of other Asian endoscopists and 75.0 % of North American endoscopists had not been
using any classification at all. As expected, 76 % of Japanese and Korean endoscopists
would not biopsy to confirm AG or IM while 81.4 % of the rest of the endoscopists
would biopsy. However, the number of biopsies to be taken significantly varied among
the participants. For non-Japanese and non-Korean endoscopists, 60.0 % of them would
take four to five biopsies during upper endoscopy, following the updated Sydney system
(or modified Sydney system omitting the incisura biopsy), while only 2.9 % of Japanese
and Korean endoscopists would take similar number of biopsies.
Table 2
Current practice in assessment and reporting of AG & IM.
|
All (n = 249)
|
Japan & Korea (n = 34)
|
Rest of world (n = 215)
|
P value
|
|
Report on endoscopic finding of AG & IM
|
0.000
|
|
|
176 (70.7 %)
|
18 (52.9 %)
|
158 (73.5 %)
|
|
|
42 (16.9 %)
|
15 (44.1 %)
|
27 (12.6 %)
|
|
|
12 (4.8 %)
|
0
|
12 (5.6 %)
|
|
|
19 (7.6 %)
|
1 (2.9 %)
|
18 (8.4 %)
|
|
Classification system used for AG
|
0.000
|
|
|
57 (22.9 %)
|
32 (94.1 %)
|
25 (11.6 %)
|
|
|
11 (4.4 %)
|
2 (5.9 %)
|
9 (4.2 %)
|
|
|
81 (32.5 %)
|
0
|
81 (37.7 %)
|
|
|
19 (7.6 %)
|
0
|
19 (8.8 %)
|
|
|
81 (32.5 %)
|
0
|
81 (37.7 %)
|
|
Report on distribution of IM
|
0.001
|
|
|
151 (60.6 %)
|
12 (35.3 %)
|
142 (64.7 %)
|
|
|
54 (21.7 %)
|
15 (44.1 %)
|
39 (18.1 %)
|
|
|
44 (17.7 %)
|
7 (20.6 %)
|
37 (17.2 %)
|
|
Biopsy for diagnosis of AG or IM
|
0.000
|
|
|
178 (71.5 %)
|
3 (8.8 %)
|
175 (81.4 %)
|
|
|
34 (13.7 %)
|
26 (76.5 %)
|
8 (3.7 %)
|
|
|
37 (14.9 %)
|
5 (14.7 %)
|
32 (14.9 %)
|
|
Number of biopsies taken
|
0.000
|
|
|
30 (12.0 %)
|
24 (74.2 %)
|
6 (2.8 %)
|
|
|
22 (8.8 %)
|
4 (12.9 %)
|
18 (8.4 %)
|
|
|
44 (17.7 %)
|
4 (12.9 %)
|
40 (18.6 %)
|
|
|
23 (9.2 %)
|
1 (2.9 %)
|
22 (10.2 %)
|
|
|
47 (18.9 %)
|
0
|
47 (21.9 %)
|
|
|
83 (33.3 %)
|
1 (2.9 %)
|
82 (38.1 %)
|
AG, atrophic gastritis; IM, intestinal metaplasia; OLGA, Operative Link for Gastritis
Assessment
Overall accuracy (SD) in endoscopic diagnosis of AG & IM in the image test was 84.5 ± 14.8 %
and 80.7 ± 20.8 %, respectively ([Table 3]). The diagnostic accuracy for both AG and IM was significantly higher among Japanese
and Korean endoscopists compared to the others (AG, 92.7 ± 13.1 % vs. 83.2 ± 14.7 %,
P = 0.001; IM, 91.1 ± 14.9 % vs 79.0 ± 21.2 %, P = 0.002). Endoscopic assessment of the severity of AG & IM was less accurate at 62.5 ± 17.6 %.
The accuracy of Japanese and Korean endoscopists was still significantly higher (73.5 ± 20.8 %
vs 60.8 ± 16.4 %, P = 0.000).
Table 3
Accuracy in endoscopic diagnosis and severity assessment of AG and IM.
|
Overall
|
Japan and Korea
|
Others
|
P value
|
|
Diagnosis of AG
|
84.5 ± 14.8 %
|
92.7 ± 13.1 %
|
83.2 ± 14.7 %
|
0.001
|
|
Diagnosis of IM
|
80.7 ± 20.8 %
|
91.1 ± 14.9 %
|
79.0 ± 21.2 %
|
0.002
|
|
Severity of AG & IM
|
62.5 ± 17.6 %
|
73.5 ± 20.8 %
|
60.8 ± 16.4 %
|
0.000
|
AG, atrophic gastritis; IM, intestinal metaplasia.
Discussion
In the present study, we demonstrated that there is a wide variety in practices around
the globe in detection and grading of the severity of AG and IM. Because these conditions
are established as known precursors to gastric cancer, it is desirable to develop
a more structured diagnostic approach.
Appreciating the degree of atrophy and intestinal metaplasia may help predict the
risk of gastric cancer in a patient. Traditionally, the western endoscopic community
has been using the OLGA and operative link on gastric intestinal metaplasia assessment
(OLGIM) systems to stage the severity of AG and IM, as pathological assessment is
still considered the gold standard [3]. In many Asian countries, endoscopy-based assessment such as the Kyoto and Kimura-Takemoto
classifications is used. The results of our study echoed the difference in practice.
In addition, it also showed that one-third of endoscopists did not routinely report
on the severity of AG. That may reflect a current lack of systematic education for
these conditions in different parts of the world.
There are a few theoretical benefits of real-time endoscopic assessment of AG and
IM. First of all, immediate risk stratification could be performed, allowing the endoscopist
to perform more prolonged, vigilant examination of the stomach if significant AG or
IM is identified. Second, it may reduce the number of biopsies required for each endoscopic
procedure, thus reducing the cost and potential complication from biopsies. Third,
the Sydney classification of gastritis was based on random biopsies for determination
of histological atrophy and intestinal metaplasia [11]. It may induce sampling error as the pattern of AG and IM are often patchy inside
the stomach.
Kono et al previously reported a good correlation between endoscopic assessment of
GA using the Kimura Takemoto classification and histopathological assessment in a
mixed group of Japanese and western population [12]. Moreover, use of image-enhanced endoscopy has significantly improved endoscopic
detection of premalignant gastric conditions. Features such as the light blue crest
(LBC) sign and white opaque substances (WOS) are high accurate in diagnosing IM [13]
[14]. It may also help to map the severity of IM throughout the stomach. Recently, a
new Endoscopic grading of gastric intestinal metaplasia (EGGIM) classification based
on endoscopic assessment of IM has been reported [8]. The endoscopy- based “EGGIM” system has been found to have excellent correlation
with the OLGIM histology-based system. More confirmatory studies are required but
the current trend favors use of endoscopy-based assessment as a replacement for multiple
gastric biopsies.
The accuracy of endoscopy-based assessment is highly dependent on the experience of
the endoscopist. In our study, the ability to accurately diagnose and map premalignant
gastric conditions varied between continents. Japanese endoscopists have been using
endoscopic grading methods for a long time, thus their performance is understandably
better than the rest of the world. It is essential that all endoscopists performing
gastroscopy are equipped with adequate techniques and knowledge for assessing the
severity of AG and IM before worldwide adoption of an endoscopic classification system
for risk stratification.
Our study had several limitations. First, the number of endoscopists recruited was
small and most were connected to university-affiliated or specialist cancer hospitals.
The real-world situation may be even more variable than was demonstrated in the study.
Second, in the image test, the endoscopic cases were chosen and validated by two endoscopists
with vast experience in endoscopic recognition of GA and IM, but biopsy was not taken
for confirmation, based on local practice. Although the most representative still
images were provided in the image test, the lack of a continuous video may also have
limited the participants’ ability to assess the extent of AG and IM.
Conclusion
In conclusion, there is currently a wide variety of practice in diagnosing and grading
GA and IM. Endoscopic evaluation and mapping of premalignant gastric conditions has
not been widely adopted, and the diagnostic ability of endoscopists should be further
improved through structured training and education.
